317 MS received 13 January 2006; revised 3 April 2006
Abstract. Mn doped SnOx thin films have been fabricated by extended annealing of Mn/SnO2 bilayers at 200°°C in air for 110 h. The dopant concentration was varied by controlling the thickness of the metal layer.
The overall thickness of the film was 115 nm with dopant concentration between 0 and 30 wt% of Mn. The films exhibit nanocrystalline size (10–20 nm) and presence of both SnO and SnO2. The highest transmission observed in the films was 75% and the band gap varied between 2⋅⋅7 and 3⋅⋅4 eV. Significantly, it was observed that at a dopant concentration of ~4 wt% the transmission in the films reached a minimum accompanied by a decrease in the optical band gap. At the same value of dopant concentration the resistivity also reached a peak. This behaviour appears to be a consequence of valence fluctuation in Sn between the 2+ and 4+ states.
The transparent conductivity behaviour fits into a model that attributes it to the presence of Sn interstitials rather than oxygen vacancies alone in the presence of Sn2+.
Keywords. Tin oxide; transparent conductors; thin films.
1. Introduction
Transparent conducting oxides have been the subject of research interest over a number of years (Gordon 2000;
Kikuchi et al 2002; Man-Soo et al 2003; Matsubara et al 2003; Wohlmuth and Adesida 2005). SnO2 based sys- tems, in particular, have been the focus of many of these investigations. SnO2 evinces interest because it is a natu- rally non-stoichiometric prototypical transparent conducting oxide. It has a high band gap of almost 4 eV, plasma fre- quency in the IR region and, when suitably doped, can be used both as a p-type and n-type semiconductor. It cry- stallizes in the tetragonal rutile type of structure, D144h (P42/mnm) with two Sn and four oxygens per unit cell.
The lattice parameters are a = b = 0⋅4737 nm, c = 0⋅3185 nm and c/a = 0⋅673.
Thin films of pure and doped SnO2 have been deposited by various physical and chemical methods (Brousse and Schleich 1996; Cirilli et al 1998; Dhere et al 1998; Mientus and Ellmer 1998; Yan et al 1998; Bauduin et al 1999;
Jae-Ho et al 1999; Shokr et al 2000; Robbins et al 2001;
Shamala et al 2004). It is evident from literature that most of the focus of interest in doped SnO2 based thin films has been on the SnO2:F and SnO2:Sb systems and there is scattered literature on dopants such as Nb (Kikuchi et al 2002), Al (Bagheri-Mohagheghi and Shokooh-Saremi 2004), Mo (Sterna and Granqvist 1994)and In (Enoki 1991).
There is not much work known on other metal doped SnO2 thin films. In this paper, therefore, we have investi- gated the transparency and conductivity in the Mn–SnO2 series of thin films as possible candidates for p-type transparent conducting thin films. Thin films were fabri- cated by thermal evaporation from a resistive source in the case of the metal and a W boat in the case of the oxide.
Thin films were characterized for transmission in the range, 190–2500 nm, resistivity using four-probe method, micro- structure by atomic force microscopy and structure using X-ray diffraction.
2. Experimental
Metal-doped SnO2 films were deposited by thermal evapo- ration on borosilicate glass substrates. A constant pressure of 5⋅0 × 10–6 Torr was maintained during the deposition process. Mn (99⋅9% purity) was first deposited on the borosilicate glass substrates and the SnO2 film was depo- sited sequentially on top of the metal films. The weight of metals in each deposition was maintained same. SnO2 powder of 99⋅9% purity and weight, 0⋅1150 g, was used in each deposition. Deposition rate of the films estimated from the thickness of the films was 2~3 Å/S. Films were then annealed in a furnace at 200°C for 110 h continuously.
The resistance of the films was measured in 12 h intervals in order to note the variation of resistance with the an- nealing time and annealing was stopped after the resi- stance saturated at a particular value.
*Author for correspondence (mgksp@uohyd.ernet.in)
Figure 1. Typical (a) EDAX and (b) SEM picture of Mn doped SnO2 thin film.
Figure 2. Optical absorption spectrum of (a) bare glass (1), pure SnO2 film (2) and Mn doped SnO2 thin film with Mn concentration of (b) 8⋅2 wt%, (c) 14 wt%, (d) 20 wt%, (e) 27⋅5 wt% and (f) 29⋅2 wt%.
The thickness of the films was measured with stylus surface profilometer (Model XP1 of Ambios Technology, USA). Structural analysis of the films were carried out using X-ray diffraction with Co Kα (λ = 1⋅7889 Å) radia- tion in a wide angled powder X-ray diffractometer (INEL Model CPS 120). The crystallite sizes, d, were determined by using Scherrer’s formula (d = kλ/βcosθ), where k is a
constant = 0⋅9, β is the FWHM and θ the angle at which the peak occurs. Chemical compositions of the films were determined by EDAX (PHILIPS XL 30 series environ- mental SEM at an accelerating voltage of 20 kV mode).
Optical transmission of the films in the range 190–2500 nm was measured by using a JASCO V 570 UV–VIS–NIR spectrophotometer. Microstructure of the films were stu-
doping had indeed taken place and the films were not composites of the metal and tin oxide. EDAX experi- ments done on these films clearly showed that the grains were typically comprised of both Mn and Sn metals. A typical EDAX spectrum and the corresponding SEM im- age are shown in figures 1(a) and (b), respectively. Secondly, optical absorption spectra were taken for the films and are shown in figure 2. From these figures it is evident that there are three features in the absorption spectra present in majority of the films. Furthermore, it is evident while the features at 1400 and 2000 nm are from the BSG sub- strate, the feature at 1⋅90 eV (650 nm) is neither due to the substrate nor is it due to SnO2 and it can be assigned to Mn2+ ions in the lattice (Mochizuki 1990). There is a shift in the wavelength at which it occurs clearly signifying a crystal field effect. Further evidence is presented in the form of X-ray diffraction patterns shown in figure 3. From this figure it is observed that the peaks in the doped films shift from their standard positions in the presence of the dopant. The shift in the lattice parameter is mainly due to the dopant occupying interstitial positions in the lattice.
Hence from the EDAX, optical absorption spectra and X-ray diffraction patterns, it is clear that the Mn2+ ions are acting as dopants in the SnO2 structure.
The X-ray diffraction patterns also indicate that the films dissociate partly into SnO and SnO2 resulting in non-stoichiometry. The amount of non-stoichiometry and therefore, the number of Sn2+ ions in the lattice is a function of the dopant concentration. There was complete absence of reflections from the metal dopant confirming that the metal particles were nano-crystalline within the detectable limits. There was also no evidence for the formation of solid-solution from the X-ray diffraction patterns. SnO2 films with up to 10 wt% dopant have been observed to exhibit the absence of dopant peaks (Shanti et al 1980).
In our study this was true even at concentration as high as 40%.
Figures 4(a) and (b) show transmission at 800 nm, the optical band gap and resistivity of the films as a function of dopant concentration. All films exhibited resistivity in the range 10–2 ohm-cm. The resistivity was dependent on dopant concentration and similar oscillatory behaviour in resistivity has been observed in the case of Sb doped SnO2 thin films prepared by spray pyrolysis (Elangovan et al
2θθ
Figure 3. XRD patterns of (a) pure SnO2 thin film and (b) Mn doped SnO2 films with Mn concentration of (a) 8⋅2 wt%, (b) 14 wt%, (c) 20 wt%, (d) 27⋅5 wt% and (e) 29⋅2 wt%.
The morphology of the films was studied by AFM. It was found that for lower dopant concentration, films were non-uniform. As the dopant concentration was increased, the films became denser and closely packed. The average particle size measured for lower concentrations was found to be 50 nm as shown in figure 5 and for higher dopant concentration it was found to be 60–70 nm. This clearly indicates that the metal dopant inhibits grain growth. The microstructural images taken for all doped samples revealed that the grain sizes in general were smaller than those for the undoped samples. They also revealed that doped films were denser with more uniform spherical grains. The films exhibit nanocrystalline size in the range of 10–20 nm as derived from X-ray diffraction patterns.
Figure 4. (a) Transmission at 800 nm as a function of dopant concentration of Mn and (b) resistivity as a function of dopant concentration of Mn. [The connecting lines are only a guide for the eye].
It is clear that the doped SnO2 films in the current study exhibit the properties of transparency and semiconductivity.
The transparency of the SnO2 matrix increased with the introduction of dopants and this was independent of dopant and accompanied by an increase in conductivity. Similar observations have been made on Sb doped SnO2 films (Shanti et al 1980; Nakanishi et al 1991; Shokr et al
Figure 5. AFM pictures of (a) pure SnO2 and (b) Mn doped SnO2 thin films.
have been reported to exhibit a peak in resistivity at ∼10 wt%
(Shanti et al 1980). The conductivity behaviour is, however, oscillatory indicating that several competing mechanisms are operating, these dominate depending on the level of dopant concentration.
Based on a local density approximation (LDA) (Cetin and Zunger 2002), it has been found that the Sn interstitial defects play a more important role in conduction than oxygen vacancies. It was proposed that tin interstitials produce a donor level inside the conduction band due to their loosely bound outer electrons which gives rise to donor ionization and conductivity. Oxygen vacancies on the other hand would tend to produce deep levels inside the band gap. Sn interstitials have a very low formation energy, therefore, they can exist in significant quantities and they are stable due to the intrinsic multivalency of Sn. It was shown that the presence of Sn interstitial lowers the for- mation energy of oxygen vacancy that in turn results in the natural oxygen deficiency and nonstoichiometry of SnO2. The absence of inter-conduction-band absorption is a consequence of a special feature of the band structure of SnO2, manifesting as a large internal gap inside the con- duction band that eliminates optical transitions in the visible range. The lowest direct optical transition corresponds to energy above 3⋅7–4⋅0 eV. The indirect transitions domi- nate as Sn2+ increases in the system and occurs within the conduction band at 2⋅7–3⋅1 eV causing transparency in the visible range. Furthermore, the conduction band absor- ption remains small, so the optical transparency is main- tained even in the presence of intrinsic defects. Another important feature of the SnO2 electronic structure is that the Fermi level lies deep inside the conduction band. As non-stoichiometry is introduced in the system, either in- trinsically or as a consequence of doping, the Fermi level is suppressed and eventually shifts to just below the con- duction band or above the valence band depending on the nature of dopant (Shanti et al 1980). Significantly, as predicted by the LDA calculations, it is seen from figures 4(a) and (b), that the minimum transmittance and the band gap occur at the same dopant concentration. Secondly, the value of the band gap is in the range 2⋅7–3⋅1 eV, which indicates the presence of Sn2+ ions. This is clearly in con- formity with the theory discussed above. It should be noted that the arguments for variation in resistivity behaviour of
The behaviour of the films has been explained within the framework of current models for the coexistence of tran- sparency and conductivity in SnO2 based systems.
Acknowledgements
The authors acknowledge support from the DST-ITPAR, DST-FIST, UPE, SAP and NPSM programs. One of the authors (RB) acknowledges financial support from the NPSM project.
References
Bagheri-Mohagheghi M M and Shokooh-Saremi M 2004 Appl.
Phys. 37 1248
Bauduin N, Hellegouarc F, Planade R, Amouroux J and Arefi- Khonsari F 1999 in Proceedings of the international sympo- sium on plasma chemistry (ISPC) (Czech Republic: Institute of Plasma Physics) Vol. IV, p. 1457
Brousse T and Schleich D M 1996 Sensors & Actuators B31 77 Cetin K and Zunger A 2002 Phys. Rev. Lett. 88 095501 Cirilli F, Kaciulis S, Mattogno G, Galolikas A, Mironas A,
Senuliene D and Setkus A 1998 Thin Solid Films 315 310 Dhere R G, Moutinho H R, Asher S, Young D, Li X, Ribelin R
and Gessert T 1998 Proc. of national centre for photovoltaics program review meeting, NREL/CP-520-25733
Elangovan E, Ramesh K and Ramamurthi K 2004 Solid State Commun. 130 523
Enoki H 1991 J. Mater. Sci. Lett. 10 970 Gordon R G 2000 MRS Bull. 25 56
Jae-Ho C, Yong-Sahm C and Dae-Seung K 1999 Thin Solid Films 349 126
Jin M, Xxiaotao H, Honglei M, Xiangang X, Yingge Y, Huang S, Zhang D and Cheng C 2002 Solid State Commun. 121 345 Kikuchi N, Kusano E, Kishio E and Kingara A 2002 Vacuum 66
365
Man-Soo H, Lee H J, Jeong H S, Seo Y W and Kwon S J 2003 Surf. & Coat. Technol. 29 171
Matsubara K, Fons P, Iwata K, Yamada A, Sakurai K, Tampo H and Niki S 2003 Thin Solid Films 431–432 369
Mientus R and Ellmer K 1998 Surf. & Coat. Technol. 98 1267 Mochizuki S 1990 J. Phys.: Condens. Matter 2 7225
Nakanishi Y, Suzuki Y, Nakamura T, Hatanaka Y, Fukuda Y, Fujisawa A and Shimaoka G 1991 Appl. Surf. Sci. 48/49 55 Pan X Q and Fu L 2001 J. Appl. Phys. 89 6048
Robbins J J, Alexander R T, Mailasu B, Yen-Jung H, Tyrone L Vincent and Wolden C A 2001 J. Vac. Sci. Technol. A19 2766 Shamala K S, Murthy L C S and Narasimha Rao K 2004 Bull.
Mater. Sci. 27 295
Shanti E, Dutta V, Banerjee A and Chopra K L 1980 J. Appl.
Phys. 51 6243
Shokr E K, Wakkad M M, Abd El-Ghanny H A and Ali H M 2000 J. Phys. Chem. Solids 61 75
Sterna B and Granquist C G 1994 J. Appl. Phys. 76 3797 Wohlmuth W and Adesida I 2005 Thin Solid Films 479 223 Yan H, Chen G H, Man W K, Wong S P and Kwok R W M
1998 Thin Solid Films 326 88
